广州市土地覆被格局异质性的尺度与等级特征

基于SPOT遥感数据源形成广州市土地覆被类型图，在中心城区选取互相垂直的南北和东西两条样带．用Fortran编程计算各样带在125m、250m、500m和1000m4个幅度下的景观多样性指数．以及Moran I、Geary C系数和半变异函数。再将样带分别自北向南和自西向东等间距划分．形成一系列面积相等的区域，计算各区域以及样带整体上的多样性指数的尺度方差。结果表明．土地覆被景观多样性在不同的研究幅度下都存在正的空间自相关性，并具有方向性。尺度方差结果显示，尺度方差与尺度和幅度都有关，随着研究尺度增大，尺度方差都呈下降之势，而随幅度增大，尺度方差并非单调变化。如125m和250m两种幅度时．方差随着尺度增大而减少。尺度方差结果进一步揭示研究样带上土地覆被存在多尺度等级结构．并且具有方向性特点．同时也反映尺度方差不失为景观异质性研究的一种有效方法。

Characteristics of Scale and Hierachical Structure of Landscape System under Different Heterogeneities of Land Cover Patterns in Guangzhou City

The pattern of land cover in Guangzhou city was translated from remote sensing image of SPOT. Two transects of north-south (N/S) and west-east (W/E) were set at the cross point which was the center of Guangzhou. Landscape diversity was selected as an index to describe the abundance of landscape. Four kinds of different extents were chosen to calculate the indices, such as 125 m, 250 m, 500 m and 1000 m. All those calculations were performed with Fortran programs under moving square windows of corresponding extents. The moving windows were sampled along the transects from west to east and from north to south with the step of a moving window size. Based on the calculation of diversity index, coefficients Moran I and Geary C and function of semi-variance were counted at different extents for correlativity of landscape diversity of urban land cover. Then, the two transects were partitioned by equal distance of the small width of the transects, 8000 m. A series of sample units with equal area were formed along the two transects. Scale variances were figured out for each unit based on the indices of landscape diversity. It was carried out at four extents. The scale variance of each transect was an average of all units in the same transect. The results revealed that there was a positively spatial autocorrelation between landscape diversities under different extents and the two transects had different spatial heterogeneities of landscape. Besides internal factors, variances of spatial structure of land cover in the two transects were affected by human activities directly. Different rates of factors formed different spatial heterogeneities of landscape of urban land cover. The semi-variance curve of diversity index was not different obviously with the increasing extents at N/S transect, but it was obvious at W/E transect. The rate of Co/(Co C) increased monotonously, indicating that the impact of stochastic factors on landscape diversity increased with the increase of research extent. The scale variance of landscape of land cover in the moving windows along the two transects under different extents decreased with the increase of scales, and all scale variances went to be almost similar at a large extent of 1000 m. The values of scale variances were obviously relative to spatially geometrical shape. The scale variance of landscape was bigger when spatial shape of sample units was a rectangle in contrast to the shape of a square. The spatial heterogeneity of landscape in the two transects was not only relative to the direction of transect, but also relative to the extent. The scale variance did not monotonously change from small extent to large extent, such as extents of 125 m and 250 m, the change of scale variance decreased with the increase of extent. Analysis of scale variance showed that there was a multi-scale hierachical structure of land cover, but it was different between the N/S and W/E transects. The heterogeneity was higher in the W/E transect than in the N/S transect. In a word, the scale variance and function of semi-variance are appropriate methods to study spatial heterogeneity and hierachical structure of landscape of urban land cover with changes in scale and extent.